op public repos

Blob

Date:: Sat Dec 31 17:24:08 2022 UTC
Message:: fix previous; name must be initialized otherwise if binaryfilename fails to match the line, `name' may be left pointing to free'd memory.
Actions:: History | Blame | Raw File
1 /*
2  * Copyright 1986, Larry Wall
3  *
4  * Redistribution and use in source and binary forms, with or without
5  * modification, are permitted provided that the following condition is met:
6  * 1. Redistributions of source code must retain the above copyright notice,
7  * this condition and the following disclaimer.
8  *
9  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND ANY
10  * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED
11  * WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
12  * DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE FOR
13  * ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
14  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
15  * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
16  * CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
17  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
18  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
19  * SUCH DAMAGE.
20  */
21 
22 /*
23  * Copyright (c) 2022 Omar Polo <op@openbsd.org>
24  *
25  * Permission to use, copy, modify, and distribute this software for any
26  * purpose with or without fee is hereby granted, provided that the above
27  * copyright notice and this permission notice appear in all copies.
28  *
29  * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
30  * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
31  * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
32  * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
33  * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
34  * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
35  * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
36  */
37 
38 #include <sys/types.h>
39 #include <sys/queue.h>
40 #include <sys/uio.h>
41 
42 #include <ctype.h>
43 #include <limits.h>
44 #include <paths.h>
45 #include <sha1.h>
46 #include <stdint.h>
47 #include <stdio.h>
48 #include <stdlib.h>
49 #include <string.h>
50 #include <unistd.h>
51 #include <imsg.h>
52 
53 #include "got_error.h"
54 #include "got_object.h"
55 
56 #include "got_lib_delta.h"
57 #include "got_lib_object.h"
58 #include "got_lib_privsep.h"
59 #include "got_lib_sha1.h"
60 
61 struct imsgbuf ibuf;
62 
63 static const struct got_error *
64 send_patch(const char *oldname, const char *newname, const char *commitid,
65     const char *blob, const int xbit, int git)
66 {
67 	struct got_imsg_patch p;
68 
69 	memset(&p, 0, sizeof(p));
70 
71 	if (oldname != NULL)
72 		strlcpy(p.old, oldname, sizeof(p.old));
73 
74 	if (newname != NULL)
75 		strlcpy(p.new, newname, sizeof(p.new));
76 
77 	if (commitid != NULL)
78 		strlcpy(p.cid, commitid, sizeof(p.cid));
79 
80 	if (blob != NULL)
81 		strlcpy(p.blob, blob, sizeof(p.blob));
82 
83 	p.xbit = xbit;
84 	p.git = git;
85 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH, 0, 0, -1, &p, sizeof(p)) == -1)
86 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH");
87 	return NULL;
88 }
89 
90 static const struct got_error *
91 send_patch_done(void)
92 {
93 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_DONE, 0, 0, -1,
94 	    NULL, 0) == -1)
95 		return got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
96 	return got_privsep_flush_imsg(&ibuf);
97 }
98 
99 /* based on fetchname from usr.bin/patch/util.c */
100 static const struct got_error *
101 filename(const char *at, char **name)
102 {
103 	char	*tmp, *t;
104 
105 	*name = NULL;
106 	if (*at == '\0')
107 		return NULL;
108 
109 	while (isspace((unsigned char)*at))
110 		at++;
111 
112 	/* files can be created or removed by diffing against /dev/null */
113 	if (!strncmp(at, _PATH_DEVNULL, sizeof(_PATH_DEVNULL) - 1))
114 		return NULL;
115 
116 	tmp = strdup(at);
117 	if (tmp == NULL)
118 		return got_error_from_errno("strdup");
119 	if ((t = strchr(tmp, '\t')) != NULL)
120 		*t = '\0';
121 	if ((t = strchr(tmp, '\n')) != NULL)
122 		*t = '\0';
123 
124 	*name = strdup(tmp);
125 	free(tmp);
126 	if (*name == NULL)
127 		return got_error_from_errno("strdup");
128 	return NULL;
129 }
130 
131 static int
132 binary_deleted(const char *line)
133 {
134 	const char *prefix = "Binary files ";
135 	const char *suffix = " and /dev/null differ\n";
136 	size_t len, d;
137 
138 	if (strncmp(line, prefix, strlen(prefix)) != 0)
139 		return 0;
140 	line += strlen(prefix);
141 
142 	len = strlen(line);
143 	if (len <= strlen(suffix))
144 		return 0;
145 	d = len - strlen(suffix);
146 	return (strcmp(line + d, suffix) == 0);
147 }
148 
149 static const struct got_error *
150 binaryfilename(const char *at, char **name)
151 {
152 	const char *suffix = " and /dev/null differ\n";
153 	size_t len, d;
154 
155 	*name = NULL;
156 
157 	len = strlen(at);
158 	if (len <= strlen(suffix))
159 		return NULL;
160 
161 	d = len - strlen(suffix);
162 	if (strcmp(at + d, suffix) != 0)
163 		return NULL;
164 
165 	*name = strndup(at, d);
166 	if (*name == NULL)
167 		return got_error_from_errno("strndup");
168 	return NULL;
169 }
170 
171 static int
172 filexbit(const char *line)
173 {
174 	char *m;
175 
176 	m = strchr(line, '(');
177 	if (m && !strncmp(m + 1, "mode ", 5))
178 		return strncmp(m + 6, "755", 3) == 0;
179 
180 	return 0;
181 }
182 
183 static const struct got_error *
184 blobid(const char *line, char **blob, int git)
185 {
186 	uint8_t digest[SHA1_DIGEST_LENGTH];
187 	size_t len;
188 
189 	*blob = NULL;
190 
191 	len = strspn(line, "0123456789abcdefABCDEF");
192 	if ((*blob = strndup(line, len)) == NULL)
193 		return got_error_from_errno("strndup");
194 
195 	if (!git && !got_parse_sha1_digest(digest, *blob)) {
196 		/* silently ignore invalid blob ids */
197 		free(*blob);
198 		*blob = NULL;
199 	}
200 	return NULL;
201 }
202 
203 static const struct got_error *
204 patch_start(int *git, char **cid, FILE *fp)
205 {
206 	const struct got_error *err = NULL;
207 	char	*line = NULL;
208 	size_t	 linesize = 0;
209 	ssize_t	 linelen;
210 
211 	*git = 0;
212 
213 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
214 		if (!strncmp(line, "diff --git ", 11)) {
215 			*git = 1;
216 			free(*cid);
217 			*cid = NULL;
218 			break;
219 		} else if (!strncmp(line, "diff ", 5)) {
220 			*git = 0;
221 			free(*cid);
222 			*cid = NULL;
223 		} else if (!strncmp(line, "commit - ", 9)) {
224 			free(*cid);
225 			err = blobid(line + 9, cid, *git);
226 			if (err)
227 				break;
228 		} else if (!strncmp(line, "--- ", 4) ||
229 		    !strncmp(line, "+++ ", 4) ||
230 		    !strncmp(line, "blob - ", 7) ||
231 		    binary_deleted(line)) {
232 			/* rewind to previous line */
233 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
234 				err = got_error_from_errno("fseeko");
235 			break;
236 		}
237 	}
238 
239 	free(line);
240 	if (ferror(fp) && err == NULL)
241 		err = got_error_from_errno("getline");
242 	if (feof(fp) && err == NULL)
243 		err = got_error(GOT_ERR_NO_PATCH);
244 	return err;
245 }
246 
247 static const struct got_error *
248 find_diff(int *done, int *next, FILE *fp, int git, const char *commitid)
249 {
250 	const struct got_error *err = NULL;
251 	char	*old = NULL, *new = NULL;
252 	char	*blob = NULL;
253 	char	*line = NULL;
254 	size_t	 linesize = 0;
255 	ssize_t	 linelen;
256 	int	 create, delete_binary = 0, rename = 0, xbit = 0;
257 
258 	*done = 0;
259 	*next = 0;
260 	while ((linelen = getline(&line, &linesize, fp)) != -1) {
261 		/*
262 		 * Ignore the Index name like GNU and larry' patch,
263 		 * we don't have to follow POSIX.
264 		 */
265 
266 		if (!strncmp(line, "--- ", 4)) {
267 			free(old);
268 			err = filename(line+4, &old);
269 		} else if (rename && !strncmp(line, "rename from ", 12)) {
270 			free(old);
271 			err = filename(line+12, &old);
272 		} else if (!strncmp(line, "+++ ", 4)) {
273 			free(new);
274 			err = filename(line+4, &new);
275 		} else if (!strncmp(line, "blob + ", 7) ||
276 		    !strncmp(line, "file + ", 7)) {
277 			xbit = filexbit(line);
278 		} else if (!git && !strncmp(line, "blob - ", 7)) {
279 			free(blob);
280 			err = blobid(line + 7, &blob, git);
281 		} else if (!strncmp(line, "Binary files ", 13)) {
282 			delete_binary = 1;
283 			free(old);
284 			err = binaryfilename(line + 13, &old);
285 		} else if (rename && !strncmp(line, "rename to ", 10)) {
286 			free(new);
287 			err = filename(line + 10, &new);
288 		} else if (git && !strncmp(line, "similarity index 100%", 21))
289 			rename = 1;
290 		else if (git && !strncmp(line, "new file mode 100", 17))
291 			xbit = strncmp(line + 17, "755", 3) == 0;
292 		else if (git && !strncmp(line, "index ", 6)) {
293 			free(blob);
294 			err = blobid(line + 6, &blob, git);
295 		} else if (!strncmp(line, "diff ", 5)) {
296 			/* rewind to previous line */
297 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
298 				err = got_error_from_errno("fseeko");
299 			*next = 1;
300 			break;
301 		}
302 
303 		if (err)
304 			break;
305 
306 		/*
307 		 * Git-style diffs with "similarity index 100%" don't
308 		 * have any hunks and ends with the "rename to foobar"
309 		 * line.
310 		 */
311 		if (rename && old != NULL && new != NULL) {
312 			*done = 1;
313 			err = send_patch(old, new, commitid,
314 			    blob, xbit, git);
315 			break;
316 		}
317 
318 		/*
319 		 * Diffs that remove binary files have no hunks.
320 		 */
321 		if (delete_binary && old != NULL) {
322 			*done = 1;
323 			err = send_patch(old, new, commitid,
324 			    blob, xbit, git);
325 			break;
326 		}
327 
328 		if (!strncmp(line, "@@ -", 4)) {
329 			create = !strncmp(line+4, "0,0", 3);
330 			if ((old == NULL && new == NULL) ||
331 			    (!create && old == NULL))
332 				err = got_error(GOT_ERR_PATCH_MALFORMED);
333 			else
334 				err = send_patch(old, new, commitid,
335 				    blob, xbit, git);
336 
337 			if (err)
338 				break;
339 
340 			/* rewind to previous line */
341 			if (fseeko(fp, -linelen, SEEK_CUR) == -1)
342 				err = got_error_from_errno("fseeko");
343 			break;
344 		}
345 	}
346 
347 	free(old);
348 	free(new);
349 	free(blob);
350 	free(line);
351 	if (ferror(fp) && err == NULL)
352 		err = got_error_from_errno("getline");
353 	if (feof(fp) && err == NULL)
354 		err = got_error(GOT_ERR_NO_PATCH);
355 	return err;
356 }
357 
358 static const struct got_error *
359 strtolnum(char **str, int *n)
360 {
361 	char		*p, c;
362 	const char	*errstr;
363 
364 	for (p = *str; isdigit((unsigned char)*p); ++p)
365 		/* nop */;
366 
367 	c = *p;
368 	*p = '\0';
369 
370 	*n = strtonum(*str, 0, INT_MAX, &errstr);
371 	if (errstr != NULL)
372 		return got_error(GOT_ERR_PATCH_MALFORMED);
373 
374 	*p = c;
375 	*str = p;
376 	return NULL;
377 }
378 
379 static const struct got_error *
380 parse_hdr(char *s, int *done, struct got_imsg_patch_hunk *hdr)
381 {
382 	static const struct got_error *err = NULL;
383 
384 	if (strncmp(s, "@@ -", 4)) {
385 		*done = 1;
386 		return NULL;
387 	}
388 
389 	s += 4;
390 	if (!*s)
391 		return NULL;
392 	err = strtolnum(&s, &hdr->oldfrom);
393 	if (err)
394 		return err;
395 	if (*s == ',') {
396 		s++;
397 		err = strtolnum(&s, &hdr->oldlines);
398 		if (err)
399 			return err;
400 	} else
401 		hdr->oldlines = 1;
402 
403 	if (*s == ' ')
404 		s++;
405 
406 	if (*s != '+' || !*++s)
407 		return got_error(GOT_ERR_PATCH_MALFORMED);
408 	err = strtolnum(&s, &hdr->newfrom);
409 	if (err)
410 		return err;
411 	if (*s == ',') {
412 		s++;
413 		err = strtolnum(&s, &hdr->newlines);
414 		if (err)
415 			return err;
416 	} else
417 		hdr->newlines = 1;
418 
419 	if (*s == ' ')
420 		s++;
421 
422 	if (*s != '@')
423 		return got_error(GOT_ERR_PATCH_MALFORMED);
424 
425 	if (hdr->oldfrom >= INT_MAX - hdr->oldlines ||
426 	    hdr->newfrom >= INT_MAX - hdr->newlines ||
427 	    /* not so sure about this one */
428 	    hdr->oldlines >= INT_MAX - hdr->newlines - 1 ||
429 	    (hdr->oldlines == 0 && hdr->newlines == 0))
430 		return got_error(GOT_ERR_PATCH_MALFORMED);
431 
432 	if (hdr->oldlines == 0) {
433 		/* larry says to "do append rather than insert"; I don't
434 		 * quite get it, but i trust him.
435 		 */
436 		hdr->oldfrom++;
437 	}
438 
439 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_HUNK, 0, 0, -1,
440 	    hdr, sizeof(*hdr)) == -1)
441 		return got_error_from_errno(
442 		    "imsg_compose GOT_IMSG_PATCH_HUNK");
443 	return NULL;
444 }
445 
446 static const struct got_error *
447 send_line(const char *line)
448 {
449 	static const struct got_error *err = NULL;
450 	char *p = NULL;
451 
452 	if (*line != '+' && *line != '-' && *line != ' ' && *line != '\\') {
453 		if (asprintf(&p, " %s", line) == -1)
454 			return got_error_from_errno("asprintf");
455 		line = p;
456 	}
457 
458 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_LINE, 0, 0, -1,
459 	    line, strlen(line) + 1) == -1)
460 		err = got_error_from_errno(
461 		    "imsg_compose GOT_IMSG_PATCH_LINE");
462 
463 	free(p);
464 	return err;
465 }
466 
467 static const struct got_error *
468 peek_special_line(FILE *fp)
469 {
470 	const struct got_error *err;
471 	int ch;
472 
473 	ch = fgetc(fp);
474 	if (ch != EOF && ch != '\\') {
475 		ungetc(ch, fp);
476 		return NULL;
477 	}
478 
479 	if (ch == '\\') {
480 		err = send_line("\\");
481 		if (err)
482 			return err;
483 	}
484 
485 	while (ch != EOF && ch != '\n')
486 		ch = fgetc(fp);
487 
488 	if (ch != EOF || feof(fp))
489 		return NULL;
490 	return got_error(GOT_ERR_IO);
491 }
492 
493 static const struct got_error *
494 parse_hunk(FILE *fp, int *done)
495 {
496 	static const struct got_error *err = NULL;
497 	struct got_imsg_patch_hunk hdr;
498 	char	*line = NULL, ch;
499 	size_t	 linesize = 0;
500 	ssize_t	 linelen;
501 	int	 leftold, leftnew;
502 
503 	linelen = getline(&line, &linesize, fp);
504 	if (linelen == -1) {
505 		*done = 1;
506 		goto done;
507 	}
508 
509 	err = parse_hdr(line, done, &hdr);
510 	if (err)
511 		goto done;
512 	if (*done) {
513 		if (fseeko(fp, -linelen, SEEK_CUR) == -1)
514 			err = got_error_from_errno("fseeko");
515 		goto done;
516 	}
517 
518 	leftold = hdr.oldlines;
519 	leftnew = hdr.newlines;
520 
521 	while (leftold > 0 || leftnew > 0) {
522 		linelen = getline(&line, &linesize, fp);
523 		if (linelen == -1) {
524 			if (ferror(fp)) {
525 				err = got_error_from_errno("getline");
526 				goto done;
527 			}
528 
529 			/* trailing newlines may be chopped */
530 			if (leftold < 3 && leftnew < 3) {
531 				*done = 1;
532 				break;
533 			}
534 
535 			err = got_error(GOT_ERR_PATCH_TRUNCATED);
536 			goto done;
537 		}
538 		if (line[linelen - 1] == '\n')
539 			line[linelen - 1] = '\0';
540 
541 		/* usr.bin/patch allows '=' as context char */
542 		if (*line == '=')
543 			*line = ' ';
544 
545 		ch = *line;
546 		if (ch == '\t' || ch == '\0')
547 			ch = ' ';	/* the space got eaten */
548 
549 		switch (ch) {
550 		case '-':
551 			leftold--;
552 			break;
553 		case ' ':
554 			leftold--;
555 			leftnew--;
556 			break;
557 		case '+':
558 			leftnew--;
559 			break;
560 		default:
561 			err = got_error(GOT_ERR_PATCH_MALFORMED);
562 			goto done;
563 		}
564 
565 		if (leftold < 0 || leftnew < 0) {
566 			err = got_error(GOT_ERR_PATCH_MALFORMED);
567 			goto done;
568 		}
569 
570 		err = send_line(line);
571 		if (err)
572 			goto done;
573 
574 		if ((ch == '-' && leftold == 0) ||
575 		    (ch == '+' && leftnew == 0)) {
576 			err = peek_special_line(fp);
577 			if (err)
578 				goto done;
579 		}
580 	}
581 
582 done:
583 	free(line);
584 	return err;
585 }
586 
587 static const struct got_error *
588 read_patch(struct imsgbuf *ibuf, int fd)
589 {
590 	const struct got_error *err = NULL;
591 	FILE *fp;
592 	int git, patch_found = 0;
593 	char *cid = NULL;
594 
595 	if ((fp = fdopen(fd, "r")) == NULL) {
596 		err = got_error_from_errno("fdopen");
597 		close(fd);
598 		return err;
599 	}
600 
601 	while ((err = patch_start(&git, &cid, fp)) == NULL) {
602 		int done, next;
603 
604 		err = find_diff(&done, &next, fp, git, cid);
605 		if (err)
606 			goto done;
607 		if (next)
608 			continue;
609 
610 		patch_found = 1;
611 
612 		while (!done) {
613 			err = parse_hunk(fp, &done);
614 			if (err)
615 				goto done;
616 		}
617 
618 		err = send_patch_done();
619 		if (err)
620 			goto done;
621 	}
622 
623 done:
624 	fclose(fp);
625 	free(cid);
626 
627 	/* ignore trailing gibberish */
628 	if (err != NULL && err->code == GOT_ERR_NO_PATCH && patch_found)
629 		err = NULL;
630 
631 	return err;
632 }
633 
634 int
635 main(int argc, char **argv)
636 {
637 	const struct got_error *err = NULL;
638 	struct imsg imsg;
639 #if 0
640 	static int attached;
641 	while (!attached)
642 		sleep(1);
643 #endif
644 
645 	imsg_init(&ibuf, GOT_IMSG_FD_CHILD);
646 #ifndef PROFILE
647 	/* revoke access to most system calls */
648 	if (pledge("stdio recvfd", NULL) == -1) {
649 		err = got_error_from_errno("pledge");
650 		got_privsep_send_error(&ibuf, err);
651 		return 1;
652 	}
653 #endif
654 
655 	err = got_privsep_recv_imsg(&imsg, &ibuf, 0);
656 	if (err)
657 		goto done;
658 	if (imsg.hdr.type != GOT_IMSG_PATCH_FILE || imsg.fd == -1) {
659 		err = got_error(GOT_ERR_PRIVSEP_MSG);
660 		goto done;
661 	}
662 
663 	err = read_patch(&ibuf, imsg.fd);
664 	if (err)
665 		goto done;
666 	if (imsg_compose(&ibuf, GOT_IMSG_PATCH_EOF, 0, 0, -1,
667 	    NULL, 0) == -1) {
668 		err = got_error_from_errno("imsg_compose GOT_IMSG_PATCH_EOF");
669 		goto done;
670 	}
671 	err = got_privsep_flush_imsg(&ibuf);
672 done:
673 	imsg_free(&imsg);
674 	if (err != NULL) {
675 		got_privsep_send_error(&ibuf, err);
676 		err = NULL;
677 	}
678 	if (close(GOT_IMSG_FD_CHILD) == -1 && err == NULL)
679 		err = got_error_from_errno("close");
680 	if (err && err->code != GOT_ERR_PRIVSEP_PIPE)
681 		fprintf(stderr, "%s: %s\n", getprogname(), err->msg);
682 	return err ? 1 : 0;
683 }